Artificial Intelligence and Robotics blog
Awesome-o
This user hasn't shared any biographical information
Posts by Awesome-o
Geminoid F android unveiled
Apr 5th
Japanese robotics researcher Hiroshi Ishiguro is continuing his quest to build androids that look and behave as close to human as possible. After creating an android copy of his young daughter, a TV celebrity and himself, earlier today he announced that he has created a new state-of-the-art android called Geminoid F.
The new android is a clone of a 20-year old female model both in appearance and behavior. It has been designed using new actuation methods that allow for a larger number of facial expressions to be displayed using less hardware. According to the press release,
The humanoid robot “Geminoid HI-1” which has previously been developed by ATR Intelligent Robotics and Communication Laboratories (ATR-IRC) has a complicated body structure with 46-degree-of-freedom. In contrast, the newly-developed Geminoid F has minimum degrees of freedom required for tele-operation and has been designed to re-create more natural facial expressions like smiling. By limiting the number of moving parts, development cost was significantly reduced. In addition, wide applications in practical scenes can be expected by adopting the appearance of a more friendly woman.
It all sounds good on paper but what does it look in real life? You can be the judge of Ishiguro’s newest android after watching it in action in the the following video.
[source]
PR2 robot folding towels
Apr 3rd
Robots have been good at keeping our floors clean and some have started working as receptionists but the household robots of our dreams are still far from real. Researchers, however, are slowly but surely making small steps towards the development of a general purpose, household robot. Pieter Abbeel and his students at the University of Berkley and in collaboration with Willow Garage have developed a new algorithm that allows a humanoid robot to fold towels, a very challenging task.
In the past, many researchers have shown how a robot can pick-up and manipulate rigid objects. The Berkley researchers have now developed a method for determining the corner points of a non-rigid object useful for grasping it and then using the PR2 showed how it works in having the robot fold towels left on a table.
This is how it works,
The robot begins by picking up a randomly dropped towel from a table, goes through a sequence of vision-based re-grasps and manipulations—partially in the air, partially on the table—and finally stacks the folded towel in a target location. The reliability and robustness of our algorithm enables for the first time a robot with general purpose manipulators to reliably and fully-autonomously fold previously unseen towels, demonstrating success on all 50 out of 50 single-towel trials as well as on a pile of 5 towels.
Sounds good doesn’t it? Other than the fact that it takes the robot a considerable amount of time to complete its task (on average it takes the PR2 nearly 1500 seconds to fold a towel) this is one of the coolest robot manipulation demos I have seen recently. If they can get the PR2 to wash my dishes, do my laundry, vacuum my floors, cook my dinner, pick-up my groceries, and walk my dog, then I will be a very happy man (I would also have more time to update this blog 
).
You can watch the robot in action in the video below.
Robots going to the dentist
Apr 1st
To clarify the title, no, our robot friends have not suddenly developed bad teeth that need to be fixed. Nor am I aware of many models that actually have teeth, but that’s a different story.
The Showa University Department of Orthodonics, School of Dentistry, has collaborated with Tmsuk to develop Hanako, a humanoid robot that can be used by dental students as practice for the real thing. Unlike a lifeless dummy, however, Hanako offers certain challenges that must be overcome by any student wishing to make his living drilling into people’s mouths.
The robot can understand and follow the surgeon’s instructions, such as when to close or open its mouth, but much like a real person, it does not stand perfectly still while being operated on. At any moment, the robot can discharge saliva, shift, sneeze or gag, forcing the student to react to the situation as best as he or she can. These motions are not entirely random, as touch sensors in Hanako’s mouth enable it to detect and react to improper application of the dental instruments, such as applying too much pressure.
There have been practice-robots in the past, but this one comes with the complete package, including movements of its jaw, eyelids, tongue or neck, especially when simulating how a human would gag. This is possibly as realistic as this type of robot can get, so we might have a rare case where one can say the model has been perfected. Not that it’s anything ground breaking, but for anyone who has had to sit on a dentist’s chair, this extra step taken to prepare students will be greatly appreciated.
The video below (in Japanese) shows a student getting to work on Hanako, and demonstrates how the robot will jerk its head if it reacts negatively to the dentist’s actions.
A model Turing machine
Mar 28th
If you don’t know what a Turing machine is but would like to after watching the below video of Mike’s magnificent device then check out the information here and/or here. You can also get additional information about the hardware, software, and example programs that Mike’s device can execute then go here.
Leading Cody by the hand
Mar 25th
![A.I. - Artificial Intelligence [Blu-ray]](http://ws.amazon.com/widgets/q?MarketPlace=US&ServiceVersion=20070822&ID=AsinImage&WS=1&Format=_SL160_&ASIN=B001993Y22&tag=artifiintella-20)
What makes this robot interesting is that it has a DPI (direct physical interface), which means that besides being operated by a playstation-like controller, simple human touch can also give it commands. It is pretty much how you would interact with a human, as you can simply take Cody’s arms and reposition it to the place you want it to be.
To be more specific, the robot possesses force sensor readings that measure how much pressure and at what angle your touch acts upon it. Pulling its arms towards or away from the robot also influences the angles of the shoulder joints, telling it where to move. At first it may seem like a blunt way of commanding a robot, but tests have shown that this direct physical interface actually works better than the remote control. Looks like there’s something to “the human touch” after all.
In design, Cody possesses little but a headless upper torso and arms, though for now it doesn’t need much else. The Mecanum wheels with which it moves seem pretty sturdy and capable of taking it where it needs to be, but everything else it accomplishes with its arms. Watching the lady leading it by the hand in the video reminds one a little bit of a child or a blind person, but appearances can be deceiving. At this point, the robot is still open-sourced, so it will be interesting to see what else it can learn.
In the video below, you can watch the robot’s Direct Physical Interface in action.
Virtobot performs digital autopsies
Mar 23rd
Few things are more grim than seeing a dead body sliced up on an operating table, but autopsy is a necessary procedure that gives us infinite information about how a person died and about the human body itself. Or, at least, it was necessary up until this point.
The “Virtobot” is a new creation by the University of Bern, Switzerland, which is able to do virtual autopsies without as much as touching the deceased. The robot applies MRI technology and advanced computer topography to scan the body literally from head to toe. It creates digital copies, which can be stored and studied whenever the information is needed.
The Swiss National Science Foundation further explained how Virtobot works, by revealing that it scans the body with a light bar, and records a high-def video of the body contours it examines. The texture of the skin is also recorded in this process. What makes this method so useful is that it allows forensic scientists to view a three-dimensional representation of the body, which can be examined both externally and internally from different angles on a screen.
The video shown at the end of this article really provides a clear picture of how Virtobot operates, and one can see just how useful it is going to be. The robotic arm that swirls around the body taking video and stills is pretty much how you would picture it in the first place.
Before one gets too excited, though, the technology still has some way to go before it is taken as more trustworthy than physical autopsies. While it can be admitted by the courts as evidence, it must still be accompanied by the standard autopsy results. All great things take time, I guess, but hopefully one day soon we will be able to rely on the Virtobot entirely.
[source]
First ever chess playing computer program
Mar 22nd
Programming computers to play games has been central to Artificial Intelligence research since its early days back in the fifties. By now, most of us have heard of IBM’s Deep Blue supercomputer defeating world chess champion Garry Kasparov in the game of chess and becoming the first ever computer to defeat man in this very complex game. Less than 3 years ago, we wrote about Jonathan Schaeffer’s unbeatable checkers program known as Chinook; it took the researchers in Schaeffer’s group 18 years to explore enough of the search space to solve the game of checkers. The same group is now trying to develop expert poker playing programs and recently they organized the first ever man-machine poker challenge (man was the winner of this first encounter.)
However, few people know about the first ever chess playing program.
This program was designed by no other than the famous British mathematician and computer pioneer Alan Turing. Starting in 1948, Turing wrote the program in collaboration with D.G. Champernowne, a former undergraduate colleague of his. Four years later in 1952, when it was time to have a match against a human opponent, Turing had no access to a computer fast enough to execute his state-of-the-art program and so he did what every one of us would do in the same situation; he simulated the program using pen and paper requiring half an hour of computation for each move!
The game was won by the human participant Alick Glennie who was another of Turing’s colleagues. Alick defeated the computer in 29 moves playing the black pieces.
If you are a chess enthusiast, you can check out the full game here. The image at the top of this post shows the board at its conclusion.
Robosoft’s Kompai to care for the elderly
Mar 18th
Japan and Korea have often been the leading countries when it comes to developing robots for the elderly, mostly due to their aging populations, but a French company called Robosoft has dipped its toes in the field of elderly care as well (remember that another French company produces the Nao humanoid robot). The robot “Kompai” can interact with its owners by speaking, or it can take commands through the touch screen mounted on its torso. What‘s cool is that it can not only repeat programmed phrases, but it can recognize and understand voices as well.
Kompai is designed to work as a personal assistance robot, helping the elderly remember things like the items on their shopping list or important dates with their doctor. It also provides phone and video communication, through which its owner can contact whoever he or she wants. Besides helping older people, it can also be of great assistance to the disabled, who need help in their daily routines.
There might be those that think this is nothing special, and it’s true that this robot isn’t exactly the most advanced and capable piece of machinery the A.I field has to offer. But take a look at the video, and see just how Kompai helps an elderly person. For the most part, it seems to understand and obey commands very well, for example knowing exactly what to ask in a situation where its owner is not feeling too well. It seems to be lacking in the ability to form grammatically correct sentences, and sometimes it sounds a bit confusing, but it gets it main message across.
One of the best things about Kompai is that it is very easy to operate, and you don’t need any special training sessions to learn how to use it—you just talk and tell it what to do and where to go.
Watch Kompai in action in the video below.
[source]
iCub: Europe’s open source robot continues to advance
Mar 16th
iCub has “grown” to resemble a three-year-old child, both in cognitive abilities and in physical appearance. The idea was to design a robot that could have a higher learning rate than any other, especially when it comes to movement. One of the robotic models can perform 53 different movements, making it a leader in human simulation.
The robot is also expected to learn how to walk and talk on its own, mimicking the progressive ways in which children develop. For that it will need to be in constant communication with humans, and learn how to interact with them. Some iCub versions can already perform the before-mentioned actions, like walking, but none have managed to match the exact growth rate of a child, which is the main goal. This means a fully-developed humanoid android, like the ones you find in Sci-fi movies, isn’t going to be ready anytime soon. But as with all great things, we’re getting there.
The only aspect they greatly need to improve I think is this supposed “toddler exterior”, which at the present time looks as far away from human as one can get. If it wasn’t for the smooth plastic head on the top, one would never guess this build of wires and steel is supposed to resemble a human. I guess junior will have to earn his skin.
NewScienstist has the video (embedded below).
Autonomous Benthic Explorer (ABE) lost at sea
Mar 15th
It has become known that a couple of days ago, the oldest operational autonomous underwater robot known as the Autonomous Benthic Explorer (ABE) was lost at sea during a mission off the coast of Chile. ABE was taken out of retirement for this last mission because the Sentry (ABE’s replacement) was kept busy at another research expedition.
According to a news release by the Woods Hole Oceanographic Institution (WHOI) makers of ABE, the vehicle was on its second dive of the mission when things went wrong.
“After a smooth launch, the dive began normally,” scientists and engineers on the ABE team reported. “ABE actively homed to its assigned position, reached the seafloor, released its descent weights, then leveled off to check its ballast. After this point, we received no more acoustic returns from the vehicle on either of its two transponders”—undersea acoustic devices that transmit and receive sound signals between vehicle and ship. The loss had nothing to do with earthquake activity off Chile, the scientists said.
ABE was equipped with several independent systems to bring it back to the surface at the end of a dive or should a fault occur. The Melville remained in the vicinity to see if ABE had resurfaced, at first searching for ABE’s strobe lights in the darkness. Researchers tried to establish radio contact with ABE in the event it had surfaced, but attempts turned up nothing.
ABE started work in 1995 and it was a truly autonomous vehicle with no tether to the research ship above. The vehicle could operate autnomously in deep waters for hours following instructions that would be uploaded on-board before the mission’s start.
It is very unfortunate that ABE was lost especially considering that it happened after the vehicle was officially retired. This was a pioneering robot and it deserved its place in the Smithsonian.
You can read more about ABE here.
Recent Comments